Task 4 - Cleaning

library(tidyverse)
library(janitor)
library(here)
library(readxl)
library(lubridate)
library(stringr)

Clean up bb_candy_2015

Clean up bb_candy_2016

bb_candy_2016_string_to_lower <- bb_candy_2016_date %>% 
  mutate(country = str_to_lower(country))

bb_candy_2016_str_replace_test <- bb_candy_2016_string_to_lower %>% 
  mutate(country = str_replace_all(country, "us", "usa"),
         country = str_replace_all(country, "usaa", "usa"),
         country = str_replace_all(country, "united states of america", "usa"),
         country = str_replace_all(country, "merica", "usa"),
         country = str_replace_all(country, "usa!", "usa"),
         country = str_replace_all(country, "u.s.", "usa"),
         country = str_replace_all(country, "usa usa usa", "usa"),
         country = str_replace_all(country, "united sates", "usa"),
         country = str_replace_all(country, "units states", "usa"),
         country = str_replace_all(country, "usa usa", "usa"),
         country = str_replace_all(country, "the yoo ess of aaayyyyyy", "usa"),
         country = str_replace_all(country, "murica", "usa"),
         country = str_replace_all(country, "the best one - usa", "usa"),
         country = str_replace_all(country, "trumpistan", "usa"),
         country = str_replace_all(country, "units state", "usa"),
         country = str_replace_all(country, "united states", "usa"),
         country = str_replace_all(country, "united state", "usa"),
         country = str_replace_all(country, "united stetes", "usa"),
         country = str_replace_all(country, "usa!!!!!", "usa"),
         country = str_replace_all(country, "usaa.", "usa"),
         country = str_replace_all(country, "usasa", "usa"),
         country = str_replace_all(country, "usa \\((i think but it's an election year so who can really tell\\))", "usa"),
         country = str_replace_all(country, "cascadia", "usa"),
         country = str_replace_all(country, "NANA.NA", "usa"),
         country = str_replace_all(country, "the republic of cascadia", "usa"),
         country = str_replace_all(country, "[0-9]", "NA"),
         country = str_replace_all(country, "united kingdom", "uk"),
         country = str_replace_all(country, "united kindon", "uk"),
         country = str_replace_all(country, "united kingdon", "uk"),
         country = str_replace_all(country, "united kindom", "uk"),
         country = str_replace_all(country, "england", "uk")
         )


bb_candy_2016_country_count_test <- bb_candy_2016_str_replace_test %>%
  distinct(country) %>% 
  summarise(n())
bb_candy_2016_country_count_test

distinct_country_remove <- bb_candy_2016_str_replace_test %>% 
  distinct(country)
view(distinct_country_remove)